Verdict
GPT-5.4 leads 16–8 across 24 shared benchmarks.
Gemini 3.1 Pro 8 · GPT-5.4 16 · higher isn't always better — check the caveats below.
Reported scores — protocols may differ. How we compare
Reasoning: Gemini 3.1 Pro leads 4–0 across 4 shared reasoning benchmarks (largest gap: SimpleQA Verified, 77.3 vs 44.83 — single evaluator).
Unknown: GPT-5.4 leads 3–1 across 4 shared unknown benchmarks (largest gap: ProofBench, 56 vs 26 — single evaluator).
Coding: GPT-5.4 leads 3–0 across 3 shared coding benchmarks (largest gap: DeepSWE, 51.77 vs 11.75 — single evaluator).
Context Learning: GPT-5.4 leads 2–0 across 2 shared context learning benchmarks (largest gap: CL-bench, 27.9 vs 20.8 — protocols undisclosed).
Math: GPT-5.4 leads 2–0 across 2 shared math benchmarks (largest gap: FrontierMath-Tier-4-v2-Private, 49 vs 26.83 — single evaluator).
Vendor claims — compare against the measured scores above. A claim is what a developer says about its own model, not an independent measurement.
Agentic / Tool Use
“Gemini 3.1 Pro was evaluated across a range of benchmarks, including reasoning, multimodal capabilities, agentic tool use, multi-lingual performance, and long-context.”
Multimodal
“a suite of highly capable, natively multimodal reasoning models”
Reasoning
“Gemini 3.1 Pro is Google's most advanced model for complex tasks.”
No vendor claims on file for GPT-5.4.
GPT-5.4 leads 16–8 across 24 shared benchmarks. Gemini 3.1 Pro leads 8 benchmarks and GPT-5.4 leads 16. Higher isn't always better — see the integrity caveats on each benchmark.
Gemini 3.1 Pro and GPT-5.4 have 24 benchmarks in common in our data — those are the only rows where a direct, apples-to-apples comparison is drawn.
Gemini 3.1 Pro leads reasoning 4–0. GPT-5.4 leads coding 3–0.